refactor: propagate bound expressions through layouts - #9126
Merged
Conversation
Merging this PR will degrade performance by 8.47%
|
| Mode | Benchmark | BASE |
HEAD |
Efficiency | |
|---|---|---|---|---|---|
| ❌ | Simulation | chunked_canonicalize_into[(1000, 50, 8, 4)] |
14.6 ms | 17.5 ms | -16.97% |
| ❌ | Simulation | chunked_canonicalize_into[(1000, 50, 8, 64)] |
17.1 ms | 20.3 ms | -16% |
| ❌ | Simulation | chunked_canonicalize_into[(1000, 50, 8, 16)] |
15.7 ms | 18.6 ms | -15.9% |
| ❌ | Simulation | chunked_canonicalize_into[(1000, 100, 16, 4)] |
20.8 ms | 24.2 ms | -13.78% |
| ❌ | Simulation | chunked_into_canonical[(1000, 50, 8, 4)] |
17.4 ms | 19.7 ms | -11.8% |
| ❌ | Simulation | chunked_canonicalize_into[(1000, 100, 16, 16)] |
24.6 ms | 27.9 ms | -11.75% |
| ❌ | Simulation | chunked_into_canonical[(1000, 50, 8, 16)] |
18.5 ms | 20.9 ms | -11.65% |
| ❌ | Simulation | chunked_into_canonical[(1000, 50, 8, 64)] |
20.1 ms | 22.7 ms | -11.41% |
| ⚡ | WallTime | cuda/bitpacked_u8/unpack/3bw[100M] |
354.3 µs | 300.5 µs | +17.9% |
| ⚡ | Simulation | chunked_varbinview_opt_into_canonical[(10, 1000)] |
6.2 ms | 5.5 ms | +13.57% |
Tip
Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.
Comparing ji/bound-expr-stack (c98bf5e) with develop (465fab3)
Footnotes
-
1 benchmark was skipped, so the baseline result was used instead. If it was deleted from the codebase, click here and archive it to remove it from the performance reports. ↩
joseph-isaacs
force-pushed
the
ji/bound-expr-stack
branch
from
August 3, 2026 12:53
1da1017 to
772e7da
Compare
joseph-isaacs
marked this pull request as ready for review
August 3, 2026 12:58
joseph-isaacs
force-pushed
the
ji/bound-expr-stack
branch
from
August 3, 2026 13:04
772e7da to
b771484
Compare
joseph-isaacs
added
changelog/break
Contributor
Polar Signals Profiling ResultsLatest Run
Previous Runs (48)
Powered by Polar Signals Cloud |
Contributor
Benchmarks: String Encoding 📖vortex / vortex-file-compressed / ms (0.998x ➖, 0↑ 0↓)
vortex / vortex-file-compressed / % (1.000x ➖, 0↑ 0↓)
|
Contributor
Benchmarks: PolarSignals Profiling 📖Vortex (geomean): 0.994x ➖ datafusion / vortex-file-compressed / ns (0.994x ➖, 0↑ 0↓)
No file size changes detected. |
Contributor
Benchmarks: TPC-H SF=1 on NVME 📖Verdict: No clear signal (low confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed / ns (0.960x ➖, 0↑ 0↓)
datafusion / vortex-compact / ns (0.957x ➖, 1↑ 0↓)
datafusion / parquet / ns (1.005x ➖, 1↑ 1↓)
duckdb / vortex-file-compressed / ns (1.001x ➖, 1↑ 0↓)
duckdb / vortex-compact / ns (1.005x ➖, 0↑ 1↓)
duckdb / parquet / ns (0.995x ➖, 1↑ 0↓)
duckdb / duckdb / ns (0.990x ➖, 0↑ 0↓)
No file size changes detected. |
Contributor
Benchmarks: FineWeb NVMe 📖Verdict: No clear signal (low confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed / ns (0.941x ➖, 2↑ 0↓)
datafusion / vortex-compact / ns (0.966x ➖, 1↑ 0↓)
datafusion / parquet / ns (0.991x ➖, 0↑ 0↓)
duckdb / vortex-file-compressed / ns (1.150x ❌, 0↑ 6↓)
duckdb / vortex-compact / ns (0.985x ➖, 1↑ 1↓)
duckdb / parquet / ns (1.003x ➖, 0↑ 0↓)
No file size changes detected. |
Contributor
Benchmarks: TPC-DS SF=1 on NVME 📖Verdict: No clear signal (low confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed / ns (0.975x ➖, 1↑ 0↓)
datafusion / vortex-compact / ns (0.982x ➖, 3↑ 1↓)
datafusion / parquet / ns (0.988x ➖, 1↑ 0↓)
duckdb / vortex-file-compressed / ns (1.025x ➖, 1↑ 11↓)
duckdb / vortex-compact / ns (1.011x ➖, 1↑ 3↓)
duckdb / parquet / ns (0.999x ➖, 3↑ 5↓)
duckdb / duckdb / ns (1.000x ➖, 2↑ 2↓)
No file size changes detected. |
Contributor
Benchmarks: Statistical and Population Genetics 📖Verdict: No clear signal (low confidence) How to read Verdict and Engines
duckdb / vortex-file-compressed / ns (1.029x ➖, 3↑ 3↓)
duckdb / vortex-compact / ns (1.062x ➖, 1↑ 4↓)
duckdb / parquet / ns (0.991x ➖, 0↑ 0↓)
No file size changes detected. |
Contributor
Benchmarks: Clickbench Sorted on NVME 📖Verdict: No clear signal (low confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed / ns (0.921x ➖, 2↑ 2↓)
datafusion / parquet / ns (1.005x ➖, 0↑ 0↓)
duckdb / vortex-file-compressed / ns (0.963x ➖, 3↑ 1↓)
duckdb / parquet / ns (0.971x ➖, 1↑ 0↓)
duckdb / duckdb / ns (0.990x ➖, 0↑ 0↓)
File Size Changes (201 files changed, -0.0% overall, 95↑ 106↓)
Totals:
|
Contributor
Benchmarks: FineWeb S3 📖Verdict: No clear signal (environment too noisy confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed / ns (0.993x ➖, 0↑ 1↓)
datafusion / vortex-compact / ns (0.841x ➖, 1↑ 0↓)
datafusion / parquet / ns (0.888x ➖, 1↑ 0↓)
duckdb / vortex-file-compressed / ns (0.993x ➖, 0↑ 0↓)
duckdb / vortex-compact / ns (1.013x ➖, 0↑ 0↓)
duckdb / parquet / ns (0.997x ➖, 0↑ 0↓)
|
Contributor
Benchmarks: Appian on NVME 📖Verdict: No clear signal (low confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed / ns (1.003x ➖, 0↑ 0↓)
datafusion / parquet / ns (1.007x ➖, 0↑ 0↓)
duckdb / vortex-file-compressed / ns (1.010x ➖, 0↑ 0↓)
duckdb / parquet / ns (1.005x ➖, 0↑ 0↓)
duckdb / duckdb / ns (1.007x ➖, 0↑ 0↓)
File Size Changes (1 files changed, -0.0% overall, 0↑ 1↓)
Totals:
|
Contributor
Benchmarks: TPC-H SF=1 on S3 📖Verdict: No clear signal (environment too noisy confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed / ns (0.941x ➖, 2↑ 1↓)
datafusion / vortex-compact / ns (1.022x ➖, 0↑ 0↓)
datafusion / parquet / ns (1.023x ➖, 0↑ 0↓)
duckdb / vortex-file-compressed / ns (0.978x ➖, 1↑ 0↓)
duckdb / vortex-compact / ns (0.999x ➖, 0↑ 0↓)
duckdb / parquet / ns (0.959x ➖, 0↑ 0↓)
|
Contributor
Benchmarks: TPC-H SF=10 on NVME 📖Verdict: No clear signal (low confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed / ns (0.987x ➖, 0↑ 0↓)
datafusion / vortex-compact / ns (0.975x ➖, 1↑ 0↓)
datafusion / parquet / ns (0.997x ➖, 0↑ 0↓)
duckdb / vortex-file-compressed / ns (1.027x ➖, 0↑ 1↓)
duckdb / vortex-compact / ns (1.005x ➖, 0↑ 1↓)
duckdb / parquet / ns (0.981x ➖, 1↑ 0↓)
duckdb / duckdb / ns (1.012x ➖, 0↑ 0↓)
No file size changes detected. |
Contributor
Benchmarks: Clickbench on NVME 📖Verdict: No clear signal (low confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed / ns (0.984x ➖, 3↑ 2↓)
datafusion / parquet / ns (0.998x ➖, 0↑ 0↓)
duckdb / vortex-file-compressed / ns (1.004x ➖, 5↑ 9↓)
duckdb / parquet / ns (1.006x ➖, 1↑ 2↓)
duckdb / duckdb / ns (1.009x ➖, 0↑ 0↓)
File Size Changes (1 files changed, -0.0% overall, 0↑ 1↓)
Totals:
|
Contributor
Benchmarks: Random Access 📖Vortex (geomean): 0.977x ➖ vortex / vortex-file-compressed / ns (0.977x ➖, 0↑ 0↓)
vortex / parquet / ns (1.001x ➖, 0↑ 0↓)
vortex / lance / ns (1.004x ➖, 0↑ 0↓)
|
Contributor
Benchmarks: TPC-H SF=10 on S3 📖Verdict: No clear signal (environment too noisy confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed / ns (0.910x ➖, 3↑ 0↓)
datafusion / vortex-compact / ns (0.937x ➖, 0↑ 0↓)
datafusion / parquet / ns (0.934x ➖, 1↑ 0↓)
duckdb / vortex-file-compressed / ns (1.077x ➖, 0↑ 2↓)
duckdb / vortex-compact / ns (1.035x ➖, 0↑ 1↓)
duckdb / parquet / ns (0.968x ➖, 0↑ 0↓)
|
Contributor
Benchmarks: Vortex queries 📖Verdict: No clear signal (low confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed / ns (1.007x ➖, 0↑ 0↓)
datafusion / parquet / ns (1.002x ➖, 0↑ 0↓)
duckdb / vortex-file-compressed / ns (1.038x ➖, 0↑ 0↓)
duckdb / parquet / ns (0.999x ➖, 0↑ 0↓)
No file size changes detected. |
Contributor
Benchmarks: Compression 📖vortex / vortex-file-compressed / ns (1.000x ➖, 0↑ 1↓)
vortex / vortex-file-compressed / bytes (0.999x ➖, 0↑ 0↓)
vortex / vortex-file-compressed / ratio (1.005x ➖, 0↑ 1↓)
vortex / parquet / ns (0.991x ➖, 0↑ 0↓)
vortex / parquet / bytes (1.000x ➖, 0↑ 0↓)
|
joseph-isaacs
force-pushed
the
ji/bound-expr-stack
branch
from
August 3, 2026 18:50
b771484 to
3e2d8b4
Compare
joseph-isaacs
force-pushed
the
ji/bound-expr-stack
branch
from
August 3, 2026 19:00
3e2d8b4 to
e1dbe37
Compare
Signed-off-by: Joe Isaacs <joe.isaacs@live.co.uk>
Signed-off-by: Joe Isaacs <joe.isaacs@live.co.uk>
joseph-isaacs
force-pushed
the
ji/bound-expr-stack
branch
from
August 4, 2026 14:03
e1dbe37 to
c85008e
Compare
joseph-isaacs
added
the
action/bench-gpu-compress
github-actions
Bot
removed
the
action/bench-gpu-compress
Contributor
GPU Compression |
robert3005
reviewed
Aug 5, 2026
robert3005
left a comment
robert3005
left a comment
Contributor
There was a problem hiding this comment.
Mechanically this is a clean, wide-but-shallow migration of LayoutReader::{pruning,filter,projection}_evaluation to BoundExpression. Tests and clippy pass locally. My findings are all in the places where the migration changed cache key semantics or quietly dropped a step; see the inline comments.
The headline one is the FileStatsLayoutReader prune cache, which I confirmed empirically now has a 0% hit rate and grows unboundedly.
A cross-cutting point behind several of the comments: after this PR the layout readers use two different cache key strategies on the same code path — ExactBoundExpr (Arc::ptr_eq identity) in StructReader, RowIdxLayoutReader and FileStatsLayoutReader, versus plain BoundExpression (structural) in DictReader and zoned::PruningState. The identity-keyed ones degrade silently to "never hits" when any caller rebinds, rather than failing. Please document on the LayoutReader trait that callers must pass a stably-bound expression, and consider binding once per reader and threading the bound expression through ScanRequest instead of rebinding at each entry point (scan/layout.rs:143, scan/multi.rs:449).
Checks run:
cargo nextest run -p vortex-layout -p vortex-array # 3310 passed, 1 skipped
cargo nextest run -p vortex-file -p vortex-scan # 155 passed
cargo clippy -p vortex-layout -p vortex-array -p vortex-file --all-targets # clean
Not run: vortex-cuda (no local CUDA), vortex-duckdb/vortex-datafusion, vortex-tui, docs, Python bindings, full workspace. The CodSpeed regressions flagged on this PR (runend decompress, patches_lookup) don't touch any changed code and look like Walltime-on-hosted-runner noise, as CodSpeed's own warning suggests.
Signed-off-by: Joe Isaacs <joe.isaacs@live.co.uk>
robert3005
approved these changes
Aug 5, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This is part of the migration to use
BoundExpressionthought vortex internals. #9120Summary
LayoutReaderevaluationBreak
LayoutReader api not take a
BoundExpressionit should be easy to make one withExpression::bind